[Python-ideas] Tulip / PEP 3156 - subprocess events

[Paul Moore]

On 17 January 2013 19:10, Guido van Rossum <guido at python.org> wrote:
> I think it would be tremendously helpful if you tried to implement the
> UNIX version of the subprocess transport. (Note that AFAIK Twisted has
> one of these too, maybe you can get some implementation ideas from
> them.)

You were right. In starting to do so, I found out that my thinking has
been solely based on a callback style of programming (users implement
protocol classes and code the relevant "data received" methods
themselves). From looking at some of the sample code, I see that this
is not really the intended usage style. At this point my head
exploded. Coroutines, what fun! I am now reading the sample code, the
section of the PEP on coroutines, and the mailing list threads on the
matter. I may be some time :-)

(The technicalities of the implementation aren't hard - it's just a
data_received type of protocol wrapper round a couple of pipes. It's
the usability and design issues that matter, and they are strongly
affected by "intended usage").

Paul

PS From the PEP, it seems that a protocol must implement the 4 methods
connection_made, data_received, eof_received and connection_lost. For
a process, which has 2 output streams involved, a single data_received
method isn't enough. I see two options - having 2 separate protocol
classes involved, or having a process protocol with a different
interface. Neither option seems obviously best, although Twisted
appears to use different protocol types for different types of
transport. How critical is the principle that there is a single type
of protocol to the PEP?